Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add LomoMao/delegate-to-deepseek-harness --skill delegate-to-deepseek-harnessgit clone --depth 1 https://github.com/LomoMao/delegate-to-deepseek-harnessWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/lomomao/delegate-to-deepseek-harness/delegate-to-deepseek-harness)<a href="https://agentmods.dev/skills/lomomao/delegate-to-deepseek-harness/delegate-to-deepseek-harness"><img src="https://agentmods.dev/badge/skills/lomomao/delegate-to-deepseek-harness/delegate-to-deepseek-harness/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/lomomao/delegate-to-deepseek-harness/delegate-to-deepseek-harness"><img src="https://agentmods.dev/badge/skills/lomomao/delegate-to-deepseek-harness/delegate-to-deepseek-harness.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00089 | $0.01148 |
| Opus 5 | $0.00044 | $0.00574 |
| Sonnet 5 | $0.00018 | $0.00230 |
| Haiku 4.5 | $0.00009 | $0.00115 |
Grade A, and why
delegate-to-deepseek-harness scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 123 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Delegate to DeepSeek Harness
Use DeepSeek Harness as a worker. Keep Codex as orchestrator, reviewer, and final authority.
Core rule
Delegate execution, not responsibility. A worker report is a claim, not proof. Review the workspace and independently verify relevant acceptance criteria before reporting success.
Verify the contract, not the whole implementation again. Delegate the work. Verify the evidence. Don't redo the work.
Decide whether to delegate
Delegate when the task is self-contained and at least one of these is true:
- implementation is mechanical, repetitive, or context-heavy;
- the task has clear acceptance criteria and a practical verification path;
- repository exploration can be framed as a concrete question;
- an isolated second implementation or diagnosis would be useful;
- independent tasks can run in separate worktrees.
Keep the task in Codex when it is trivial, primarily architecture/product judgment, dependent on rich unsummarized conversation context, privileged or irreversible, or not independently verifiable.
Choose the worker path
Use this order:
- Harness MCP if a configured server exposes
agent_runand, when needed,task_inbox/task_result. - Headless fallback with
scripts/dsh_headless.shwhen MCP is unavailable anddshexists. - If neither path is available, stop delegation and state the missing prerequisite.
Read references/setup.md only when setup or connectivity matters.
Write the worker brief
Give the worker a self-contained contract:
Objective:
<one concrete outcome>
Scope:
<allowed files/modules and explicit non-goals>
Constraints:
<compatibility, API, style, dependency, or user constraints>
Acceptance criteria:
- <observable condition>
- <observable condition>
Verification contract:
- allowed_paths: <glob list, e.g. src/parser/**>
- checks: <test/lint/typecheck/build commands to run>
- invariants: <no_new_dependencies | no_public_api_change | no_untracked_files | ...>
- limits: <max_changed_files | max_diff_lines>
- review_mode: receipt | targeted | full
Verification:
- <focused command/check if known>
Rules:
- Work only inside the supplied cwd.
- Inspect before editing.
- Preserve unrelated changes.
- Do not commit, push, publish, deploy, rotate credentials, or perform destructive external actions.
- If blocked, report the blocker instead of guessing.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 11d ago First seen · 123 lines · 89 tokens per session scan A c917d562fce2
delegate-to-deepseek-harness is a skill published in the GitHub repository LomoMao/delegate-to-deepseek-harness (2 stars, last pushed 26d ago), licensed MIT. It adds 89 tokens to every session and 1,148 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
context-optimizer
A context-management assistant for long conversations with AI agents. It helps organize the information an agent needs when conversations become large or contain conflicting details.
better-prompt
A prompt editor that turns rough instructions for AI systems into clearer, more complete prompts. It follows published OpenAI and Anthropic guidance.
git-commit
A Git assistant for reviewing local changes and creating conventional commit messages, which are standardized summaries of code changes. It can also run Git hooks and push the commit.
git-pr-review
A read-only reviewer for GitHub pull requests, which are proposed code changes submitted for review. It produces an evidence-based report about whether a pull request should be merged.
download-fulltext-pdf
A skill for downloading a research paper's complete PDF using an identifier such as a DOI, title, or BibTeX entry.
bensz-collect-bugs
A process for recording bugs caused by defects in Bensz skills. It stores a structured local report and can publish one only when the user explicitly asks.